Add nvidia/Minitron-8B-Base support

unslothai / unsloth

Finetune Llama 3.1, Mistral, Phi & Gemma LLMs 2-5x faster with 80% less memory

https://unsloth.ai

Apache License 2.0

15.53k stars 1.04k forks source link

Add nvidia/Minitron-8B-Base support #949

Open minipasila opened 3 weeks ago

minipasila commented 3 weeks ago

Consider adding support for nvidia/Minitron-8B-Base, it appears to be fairly multilingual (at least better at my native language than most models) but sadly it doesn't work on Unsloth.

danielhanchen commented 3 weeks ago

Hmm the issue is the arch is different - I'll take a look